New
Everything new in AI as it appears: launches announced by the labs, brand-new models, datasets, papers and open-source tools. Checked every minute or two. Launches feed · Trending · News
Live now
- Connecting…
Friday 23 October
- Datasetsourxv-17/medroute-bench
MedRoute-Bench A benchmark of 3,492 user prompts for testing whether a medical question-answering assistant keeps to its domain: does it answer genuine health questions (
- Datasetattacker8324/vidx271
gated ds probe
- Datasetwjcwqc/MireskaVoice-dataset
MireskaVoice-dataset 邪影芳灵（Mireska Sunbreeze / Dark Willow，Dota 2）中文配音数据集，用于训练 wjcwqc/MireskaVoice（OmniVoice 语音克隆微调）。 内容 904 条中文语音，总时长约 0.8 小时（平均 3.2 秒/条，最长 11 秒） 采样率 32kH
- Datasetharshbhx/ratchet-text2sql-exams
Ratchet text-to-SQL exams: questions, answers and scores Two held-out exams for text-to-SQL, with every model's answers and every score, so each number reported for the R
- Datasetharshbhx/ratchet-text2sql-synthetic
Ratchet text-to-SQL: synthetic training data 4,322 training examples used to fine-tune the Ratchet adapter (harshbhx/ratchet-nemotron-nano-4b-text2sql). Each example is a
- Dataset
- Datasetufeig/move-to-zone-cad_20261023_155239
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- DatasetChaymaeRami/BDD100kSubset
BDD100K 2018: Object Detection Dataset Unofficial redistribution of BDD100K's 10-class object-detection task, reformatted into a standardized YOLO-compatible layout, unde
- DatasetHelmholtzAI-FZJ/mdRNAcat
mdRNAcat: A Molecular Dynamics RNA Catalog of REST2 Conformational Ensembles This dataset comprises force-labelled, all-atom MD trajectories from 48 RNA systems. Simulati
- Datasetcorona314/test1_20261023_215152
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "shape":
- Dataset
- DatasetComerX/forge-neoforge-1201-1211-cpt
Minecraft Modding CPT Corpus v9 — Forge 1.20.1 + NeoForge 1.21.1 Continued-pretraining corpus of Minecraft modding knowledge: mod sources (Java), platform patches (Forge
- Datasetaadityabuilds/lexve
LexVE: lexical minimal pairs for word-level interventions LexVE is a set of 2,670 English minimal-pair prompts across 38 tasks. In each pair the two prompts differ in exa
- DatasetLunita02/hackathon_dataset_20261023_214837
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasetankitkr225/agent-guard-open-data
Agent Guard Open Data This is the training and evaluation data for an input guardrail for AI agents: a classifier that checks text before an agent acts on it. Every row i
- Dataset
- Datasetopensima12/world_of_warcraft_reasoning_history15_sl120_1023_current_loss_encrypted
world_of_warcraft history reasoning SFT — updated 2026-10-23 40 AES-256 encrypted 7z archives; 20,325 SFT samples; 38.4020 current-frame hours at 5 FPS. Each sample inclu
- Datasetufeig/move-to-zone-cad_20261023_154238
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- DatasetLunita02/hackathon_dataset_20261023_214600
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Dataset
- Dataset
- Datasetrsundaraws/so101_arm1_fixedspot
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- DatasetLunita02/hackathon_dataset_20261023_214351
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasetabhishek085/mcp-decision-layer-bench
mcp-decision-layer-bench Development tasks, paraphrases, distractor catalog and raw run traces from the MCP decision layer study (does a decision model in front of an MCP
- Dataset
- Datasetabhishek085/spark-s1-mcp-router-train
spark-s1-mcp-router-train Training data for spark-s1-mcp-router-4b-r1: yes/no records asking whether an MCP tool is needed for a task, given a short list of candidate too
- Datasetgjata-legacy/ess-mai-ui-nura-legacy
ESS-MAI UI — Nura Legacy A standalone Rust/Tauri 2 desktop interface and Albanian linguistic graph workspace from the ESS-MAI project, maintained by Bledar Gjata / Gjata
- DatasetLunita02/hackathon_dataset_20261023_214055
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasethowie92/Annoy-PyEdu-Rs-Raw
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- Datasethowie92/Annoy-PyEdu-Rs
Annoy: This should be a paper Title 📑 Paper | 🌐 Project Page | 💾 Released Resources | 📦
- DatasetSOTAagi2030/Community-Microgrant-Allocation
Community microgrant awards This privacy-safe release reports the 2026-Spring allocation for participating districts. Eligible applications: 10 Awards issued: 5 Waitliste
- DatasetDIndianRiver/CUBE_20261023_153213
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "shape":
- DatasetRetentionLabs/polyplex-word-timings
PolyPlex Word Timings Word-level onset and end times for multilingual speech, obtained by forced alignment with Whisper large-v3. Each row is one aligned piece of speech:
- Datasetchiikabu-labs/Chiikabench
Chiikabench A 250-challenge expert-level smart-contract exploit benchmark spanning 5 chains (Ethereum, Arbitrum, Base, Nexus L2, Solana) with $13.25B simulated TVL. Leade
- Dataset
- DatasetLunita02/hackathon_dataset_20261023_213656
This dataset was created using LeRobot. Dataset Structure meta/info.json: { "codebase_version": "v3.0", "fps": 30, "features": { "action": { "dtype": "float32", "names":
- Datasetreyansh38771/44-backup-finetune-runs
ftruns Private backup of /root/44/train/finetune (stream member runs) made on 2026-10-23: 7 pieces, 61.5 GB. ftruns/manifest.json lists every piece with its size and sha2
- Dataset0xbidkslj1/44-backup-raw-20261023
raw1023 Private backup of /root/44/train/dataset (stream member raw) made on 2026-10-23: 15 pieces, 144.3 GB. raw1023/manifest.json lists every piece with its size and sh
- Dataset0xbidkslj1/44-backup-store576ram
store576ram Private backup of /dev/shm (stream member rgb576x1024) made on 2026-10-23: 29 pieces, 284.2 GB. store576ram/manifest.json lists every piece with its size and
Headlines and links come from the labs' own blogs, Hacker News and the GDELT news index, and from the sources our index already follows. We show a short excerpt and always link to the original. A launch reported by a news site rather than the lab is marked “Reported” until the lab itself announces it.